Evaluating the Influence of R3 Treatments on Fishing License Sales in Pennsylvania

Evaluating the Influence of R3 Treatments on Fishing License Sales in Pennsylvania Prepared for the: Pennsylvania Fish and Boat Commission Produced by: PO Box 6435 Fernandina Beach, FL 32035 Tel (904) 277-9765 May 25, 2018

Executive Summary The Pennsylvania Fish and Boat Commission has implemented several different programs (referred to as treatments in this report) aimed at recruiting, reactivating, and retaining anglers, with some programs dating back as far as 2008. Treatment participation data have been analyzed to determine what impact the treatments have had on participants license buying behavior. These participating customers are referred to as treatment groups for the purposes of this report. Each treatment had specific goals and targeted different demographic groups. Altogether, three measures of effectiveness stood out across various treatments: 1. The ability to recruit and reactivate anglers at a rate higher than the natural churn rate among the treatment group; 2. The ability to increase retention among anglers, thereby decreasing the rate of churn, and; 3. The ability to influence long-term buying patterns among the treatment group. These measures were emphasized at different rates across the various treatments, some only seeking to retain anglers, while other treatments sought to be successful by all these measures. However, each treatment shared the same overarching goal: to sustainably increase the sales of fishing licenses to anglers in Pennsylvania. Each of the treatments analyzed in this report can be categorized as either active or passive. Active treatments, such as the Family Fishing and Fly Fishing programs, are hands-on and interactive, placing potential anglers in direct contact with PFBC staff and coordinators. Conversely, passive treatments, such as the Angler Awards and Contests treatments, do not require attendance at an event, do not have a designated time or place, and individuals have limited direct contact with PFBC. Though the active, hands-on programs are more intense in their ability to influence participants, the passive, hands-off programs can more easily reach a much larger group of participants. Two treatments were able to produce rates of reactivation and recruitment that kept pace with the natural churn rate of the treatment group: Angler Awards and the Fly Fishing Program, both of which were passive. These treatments have achieved the goals of measure #1 and can be considered an option to not just stave off the loss of anglers, but possibly to increase the population of anglers. The results show that, in most cases, individuals within the treatment groups had lower churn rates and relatively higher rates of license purchasing in the years following treatment when compared to those who did not participate. This suggests that most of the treatments accomplished the goal of measure #3 to some degree. The major exception to this is the treatment of selling discounted licenses late in the season. In this treatment, we actually see a possible reduction in the likeliness of treated individuals to purchase in the next year. Treatment individuals are slightly more likely to purchase in the year following their treatment than in the year preceding their treatment. Each treatment group shows an increase in license purchasing leading up to the treatment year. Then, in the years after the treatment there is a much higher purchase rate than preceding the treatment. Treatment individuals license purchasing peaks during the year of treatment, and diminishes afterward, but at a slower rate than that of the un-treated individuals. ii

Many of the treatment programs appealed to demographic groups not well-represented within the current PALS customer base. The Angler Awards treatment elicited participation from younger participants while the Contests, Family Fishing, and Fly Fishing Programs had high participation rates from women. One treatment targeted both younger participants as well as women: Fishing Skills Instructor Training. In analyzing the results presented here it is important to note that, while there are many reasons an individual would enter into a treatment group, all treatment groups were comprised of individuals who chose to participate in the contest, event, or program. This introduces both a self-selection and avidity bias to any comparison of treatment individuals to non-participants. These biases are important caveats in interpreting the results of this analysis. Self-selection bias means that each person who signed up for a treatment was not randomly assigned, but rather they proactively chose to participate and likely had their own motivation for participating in a treatment. Some of these people may have already decided that they were interested in becoming more active as anglers before participating in the treatment. In fact, this may have been what lead them to participate in the treatment. Thus, their subsequent license purchases cannot be attributed solely to their having participated in a specific program. Avidity bias means that more avid anglers may have been more likely to participate in these treatments. This can potentially obfuscate the results for treated customers, as more avid anglers are more likely to purchase licenses, regardless of their participation in specific programs. iii

Table of Contents Executive Summary... ii Table of Contents... iv List of Tables... vi Background...1 PALS Overview... 2 Demographic Information... 2 License Analysis... 3 Methodology...4 Data cleaning and organization... 4 Description of Reporting Tables... 4 Definitions of R3 terms used... 5 Representativeness of the final data set(s)... 6 Results...7 Treatment 1: Contests... 7 Data Overview... 7 Demographics... 8 License Status at Time of Treatment... 9 Recruitment, Reactivation, and Retention... 9 Treatment 2: Angler Awards... 12 Data Overview... 12 Demographics... 12 License Status at Time of Treatment... 13 Recruitment, Reactivation, and Retention... 14 Treatment 3: Family Fishing Program (All Types)... 17 Data Overview... 17 Demographics... 17 License Status at Time of Treatment... 18 Recruitment, Reactivation, and Retention... 18 Treatment 4: Fly Fishing Program (All Types and Levels)... 22 iv

Data Overview... 22 Demographics... 22 License Status at Time of Treatment... 23 Recruitment, Reactivation, and Retention... 24 Treatment 5: Fishing Skills Instructor Training... 28 Data Overview... 28 Demographics... 29 License Status at Time of Treatment... 29 Recruitment, Reactivation, and Retention... 30 Treatment 6: Late Season Discount Licenses... 34 Data Overview... 34 Demographics... 34 License Status at Time of Treatment... 35 Recruitment, Reactivation, and Retention... 36 Comparison of Treatments... 39 Conversion Rate... 39 Reactivation and Recruitment... 40 Retention... 40 Churn... 41 Recommendations... 42 Digging Deeper... 42 Understanding Potential Lasting Effects... 42 Data Quality... 42 Conclusion... 44 v

List of Tables Table 1: Age Category (PALS)... 2 Table 2: Gender (PALS)... 2 Table 3: PALS Churn Rate (PALS)... 3 Table 1.0: Data Completeness Overview (Contests)... 7 Table 1.1: PALS Record Identification (Contests)... 8 Table 1.2: Age Category (Contests)... 8 Table 1.3: Gender (Contests)... 8 Table 1.4: License Status at Time of Treatment (Contests)... 9 Table 1.5: R3 (Contests)... 10 Table 1.6: Purchase Timing of Reactivated & Recruited Customers (Contests)... 10 Table 1.7: Churn Rate (Contests)... 11 Table 1.8: Five Year Purchasing Pattern (Contests)... 11 Table 2.0: Data Completeness Overview (Angler Awards)... 12 Table 2.1: PALS Record Identification (Angler Awards)... 12 Table 2.2: Age Category (Angler Awards)... 13 Table 2.3: Gender (Angler Awards)... 13 Table 2.4: License Status at Time of Treatment (Angler Awards)... 14 Table 2.5: R3 (Angler Awards)... 14 Table 2.6: Purchase Timing of Reactivated & Recruited Customers (Angler Awards)... 14 Table 2.7: Churn Rate (Angler Awards)... 15 Table 2.8: Five Year Purchasing Pattern (Angler Awards)... 15 Table 2.8.1: Five Year Purchasing Pattern Continued (Angler Awards)... 16 Table 2.8.2: Five Year Purchasing Pattern Continued (Angler Awards)... 16 Table 2.8.3: Five Year Purchasing Pattern Continued (Angler Awards)... 16 Table 3.0: Data Completeness Overview (Family Fishing)... 17 Table 3.1: PALS Record Identification (Family Fishing)... 17 Table 3.2: Age Category (Family Fishing)... 18 Table 3.3: Gender (Family Fishing)... 18 Table 3.4: License Status at Time of Treatment (Family Fishing)... 18 Table 3.5: R3 (Family Fishing)... 19 vi

Table 3.6: Purchase Timing of Reactivated & Recruited Customers (Family Fishing)... 19 Table 3.7: Churn Rate (Family Fishing)... 20 Table 3.8: Five Year Purchasing Pattern (Family Fishing)... 20 Table 3.8.1: Five Year Purchasing Pattern Continued (Family Fishing)... 21 Table 3.8.2: Five Year Purchasing Pattern Continued (Family Fishing)... 21 Table 3.8.3: Five Year Purchasing Pattern Continued (Family Fishing)... 21 Table 4.0: Data Completeness Overview (Fly Fishing)... 22 Table 4.1: PALS Record Identification (Fly Fishing)... 22 Table 4.2: Age Category (Fly Fishing)... 23 Table 4.3: Gender (Fly Fishing)... 23 Table 4.4: License Status at Time of Treatment (Fly Fishing)... 24 Table 4.5: R3 (Fly Fishing)... 24 Table 4.6: Purchase Timing of Reactivated & Recruited Customers (Fly Fishing)... 25 Table 4.7: Churn Rate (Fly Fishing)... 25 Table 4.8: Five Year Purchasing Pattern (Fly Fishing)... 26 Table 4.8.1: Five Year Purchasing Pattern Continued (Fly Fishing)... 26 Table 4.8.2: Five Year Purchasing Pattern Continued (Fly Fishing)... 27 Table 5.0: Data Completeness Overview (Skills Instructor)... 28 Table 5.1: PALS Record Identification (Skills Instructor)... 28 Table 5.2: Age Category (Skills Instructor)... 29 Table 5.3: Gender (Skills Instructor)... 29 Table 5.4: License Status at Time of Treatment (Skills Instructor)... 30 Table 5.5: R3 (Skills Instructor)... 30 Table 5.6: Purchase Timing of Reactivated & Recruited Customers (Skills Instructor)... 31 Table 5.7: Churn Rate (Skills Instructor)... 31 Table 5.8: Five Year Purchasing Pattern (Skills Instructor)... 32 Table 5.8.1: Five Year Purchasing Pattern (Skills Instructor)... 32 Table 5.8.2: Five Year Purchasing Pattern (Skills Instructor)... 32 Table 5.8.3: Five Year Purchasing Pattern (Skills Instructor)... 33 Table 5.8.4: Five Year Purchasing Pattern (Skills Instructor)... 33 Table 6.0: Data Completeness Overview (Late Season Discount)... 34 Table 6.1: PALS Record Identification (Late Season Discount)... 34 vii

Table 6.2: Age Category (Late Season Discount)... 35 Table 6.3: Gender (Late Season Discount)... 35 Table 6.4: License Status at Time of Treatment (Late Season Discount)... 35 Table 6.5: R3 (Late Season Discount)... 36 Table 6.6: Purchase Timing of Reactivated & Recruited Customers (Late Season Discount)... 36 Table 6.7: Churn Rate (Late Season Discount)... 37 Table 6.8: Five Year Purchasing Pattern (Late Season Discount)... 37 Table 6.9: 2016 R3 Status of Those Churned in 2017... 38 Table 7.1.: Ability of Treatments to Convert Potential Purchasers... 39 Table 7.2.: License Purchase Timing of Recruited and Reactivated Anglers Across Treatments... 39 Table 7.3.: Comparison of All Treatment Reactivation Rates... 40 Table 7.4.: Comparison of All Treatment Retention Rates... 40 Table 7.5.: Comparison of All Treatment Churn Rates... 41 Table 8.1: Purchase Rates Relative to Treatment (2013)... 44 Table 8.2: Purchase Rates Relative to Treatment (2014)... 45 Table 8.3: Purchase Rates Relative to Treatment (2015)... 45 viii

Background This study was undertaken to understand how six angler programs (referred to as treatments in this report) might have aided in the recruitment, retention, and reactivation of anglers in Pennsylvania in the years from 2008 to 2017. Groups of anglers that participated in each treatment are referred to as treatment groups. The treatments were as follows: 1. Contests paper contest entry forms were completed at shows and events where PFBC had a booth; 2. Angler Awards - anglers submitted information to PFBC about fish they had caught, being awarded if they met certain requirements; 3. Family Fishing - a hands-on education program targeting families, and provided basic fishing skills as well as the opportunity to fish; 4. Fly Fishing Program - a hands-on education program that targeted families and provided basic fly fishing skills training and the opportunity to fish; 5. Fishing Skills Instructor Training - a train-the-trainer workshop that worked to certify individuals to deliver PFBC angler education programs; 6. Late Season Discount - a passive treatment that was an effort to recruit more anglers to purchase licenses by offering a discounted rate halfway through the season. There are major differences in many of these treatments. Some, like the Family Fishing, Fishing Skills Instructor, and the Fly Fishing Program, are hands-on experiences meant to engage participants in new activities. Others, like the Contests and Angler Awards, are more passive treatments meant to entice customers to participate in a very hands-off way. These were not targeted at any particular demographic group or specific type of fishing. Thus, the awards functioned as more of a retention tool than a recruitment or reactivation tool. The Late Season Discount treatment is different from the others because participation involves actually purchasing a license. Though the hands-on programs are more intense in their ability to influence participants, the hands-off programs can more easily reach a larger group of participants. We see very different types of participation in each treatment, in terms of both number of participants and demographic distribution. People have different reasons for participating in different treatments, and all participate of their own accord. Therefore, it is difficult to measure the effectiveness of an individual treatment, since subsequent license purchases cannot reasonably be attributed solely to participation in the treatment. We can, however, examine both prior purchasing behavior of participants, and the purchasing behavior of non-participants over the same time period. This allows us to get some sense of the impact that these treatments have had on participants, although any changes in license purchasing cannot be attributed entirely to the treatment itself. Methodology Page 1

PALS Overview Between 2008 and 2017 PALS cataloged 14.2M licenses sold to more than 2.8M anglers. Churn rates during this time period are relatively low, with over 60% of anglers maintaining their fishing license from year to year. Demographic Information The following summarizations of demographic information are based on all customers that purchased a PA fishing license between 2008 and 2017. The ages presented in Table 1 below represent the angler s age as of December 31, 2017, which may differ from the age at which they last purchased a license. Table 1: Age Category (PALS) Age Group Count Percent 15 or younger 64,819 2.3% 16 to 24 years 350,048 12.5% 25 to 34 years 581,080 20.7% 35 to 44 years 487,294 17.4% 45 to 54 years 514,328 18.4% 55 to 64 years 443,389 15.8% 65 to 74 years 282,767 10.1% 75 to 84 years 66,767 2.4% 85 years and older 9,955 0.4% Total 2,800,447 100% There are about three male customers for every female customer in the PALS data during the years of analysis, 2008 to 2017. This gender ratio is not consistent among all treatments. Table 2: Gender (PALS) Year Customers Percent Female 677,611 24.2% Male 2,113,736 75.5% *not reported 9,100 0.3% Total 2,800,447 100% Methodology Page 2

License Analysis Churn, as used here, denotes the proportion of anglers who purchased a license in one year, and then did not renew their license in the following year. Thus, the 2015 churn rate of 37.2% shows that 62.8% of all anglers who bought a license in 2014 were retained anglers in 2015. With the exception of 2011, all estimates of churn are below 40%. Table 3: PALS Churn Rate (PALS) Year Customers Churn 2008 747,280 n/a 2009 793,663 38.9% 2010 846,754 39.2% 2011 827,616 40.2% 2012 887,480 37.7% 2013 912,157 39.1% 2014 945,863 37.4% 2015 990,620 37.2% 2016 998,950 38.6% 2017 985,498 39.1% Methodology Page 3

Methodology Using program registration data collected by PFBC and the PALS database of all license purchases, treated individuals were matched to their PALS records to identify their license buying history (if any). Once this was accomplished, these treatment groups were examined to determine which treatments had the greatest impact in terms of increasing fishing license sales in Pennsylvania. A treatment date, corresponding with first time each person participated in a particular program or contest, was determined for each member of every treatment group. This enabled an analysis of license purchasing history for each individual both before and after the treatment took place. The license buying behavior of these treatment groups were examined against one another, as well as the greater PALS license buying population who had not entered into treatments. Data cleaning and organization As PFBC s R3 programming has developed over many years, the methods used in the collection, organization, and storage of registration data for each treatment have developed as well. Each treatment employed unique registration data collection methods, serving to support the individual treatment. However, when using these treatment registration data to identify an individual s license purchasing history, differences in data collection methods between treatments and PALS provided challenges. Treatment registration data did not include the unique customer ID s present in the PALS database. Therefore, treatment group members had to be matched to their PALS records using names and demographic information. All addresses were standardized using the CASS (Coding Accuracy Support System) provided by the U.S. Postal Service, to facilitate matching. Many individuals participated in treatments multiple times. To address this issue, an individual s earliest participation date was used to represent their treatment date. This enabled an accurate determination of timing between treatment participation and any license purchasing. For anglers matched to the PALS database, duplicates were identified as containing the same customer identification number (CID), first name, and last name. For members of the treatment group that were not matched to the PALS database, first name, last name, date of birth, and address were used to identify duplicates. Description of Reporting Tables Age and Gender: These demographic data were collected differently for every treatment. A description of the data source is included within each section of treatment results to speak to the representativeness of these tables, as some may be more representative of licensed individuals while other tables represent all individuals who entered the treatment, licensed and otherwise. License Status at Time of Treatment: these tables identify an individual's history of license purchases at the time of treatment. In order to identify those who had previously purchased licenses, individuals from the treatment group were matched with sales records from the PALS data based on a hierarchy of relationships: First, individuals treatment records were matched to PALS records based on (1) First Name, (2) Last Name, and (3) Date of Birth. Those who were not matched to PALS data based on the combination of variables above were then matched based on the following combination of variables: (1) First Name, (2) Last Name, and (3) Address. Individuals who were not matched to PALS data could fall into either of the following categories: (1) Those who had not purchased a PA fishing license between 2008 and 2017, or; Methodology Page 4

(2) Those who had incomplete data in the treatment data sets and were therefore unable to be matched with PALS data. Within the table, those who had not already purchased a license for the year in which the treatment occurred were broken out into two categories: those who had records of purchasing a PA fishing license at some point in the past, going back to 2008, and those who had no known record of fishing license purchases. Individuals from both of these categories could have gone on to purchase a license post treatment. Customer Purchase Timing: these tables describe the timing of licenses purchased by recruited and reactivated anglers. The timing of these license purchases are broken down into five categories: 1) Before the treatment date within the same license year as the treatment occurred; 2) After the treatment date, within 1 month of the treatment date; 3) 1 month after the treatment date, to less than 6 months following the treatment date; 4) 6 months after the treatment date, to less than 12 months following the treatment date; 5) 12 months after the treatment date, to 18 months following the treatment date. For Contests, Family Fishing, and Skills Instructor treatments, recruited and reactivated anglers are broken out separately in order to assess any differences between when they purchased licenses. Five Year Purchasing Pattern: these tables show the number of individuals that entered a given treatment in each year who could be identified as customers in the PALS dataset. This number appears as the total for each treatment year. This table also identifies when these customers made license purchases in the five years before and after the treatment year. Definitions of R3 terms used Because the treatments, as well as events within each treatment, took place at various times throughout any given year, R3 classifications were assigned based on the unique treatment date for each customer. Retained customers are those anglers who purchased a license within the year of their treatment, as well as the previous year. To properly classify an angler as retained, only one year of license history is required. Recruited customers are those who purchased a license either before their treatment date within the same license year, or afterwards within 18 months of their treatment date. These customers had not purchased a license in the previous five years. Five years of data is needed to accurately calculate this. Reactivated customers are those individuals who purchased a license either before their treatment date within the same license year, or afterwards within 18 months of their treatment date. These customers had purchased a license in at least one of the past five years but had not purchased a license in the year prior to the treatment. Five years of data is needed to calculate this. Within the "Reactivated" category there are two sub-definitions: Short-Term Lapse, and Long-Term Lapse. Those with a lapse of 1 to 3 years were identified as a Short-Term Lapse, while those with a lapse of 4 or 5 years were identified as a Long-Term Lapse. Individuals with a lapse of more than 5 years are identified as recruited. Methodology Page 5

Representativeness of the final data set(s) Individuals within the treatment data who were not matched to records in the PALS data could have been omitted for one of the following reasons: (1) an individual had not purchased a fishing license from PFBC within the years of analysis; (2) individuals had purchased a fishing license from PFBC within the years of analysis, but had incomplete data, disallowing license records to be matched to demographic information divulged in the treatment data. Tables within each sub-chapter of treatment results will describe the state of the data when used for analysis. Complete data is described as data that contain an individual s first name, last name, date of birth, and address; when present within the treatment data these variables are used to identify records in the PALS data, thereby illustrating a customer s history of license purchases. Observations missing date of birth, address, or both were matched to PALS data as best as possible, but the opportunity exists for false-negatives to appear. In this case, false negatives would indicate that fewer treatment participants were licensed than may have been the case. Because of data collection issues regarding treatment registrations, it is possible that the proportions in this report are not representative of the fishing license buying population. However, the assumption that errors in data collection occurred truly randomly suggests that these data, albeit a smaller sample, would be representative of the population. This assumption allows for conclusions to be drawn from the data. When interpreting these results, it is important to note that one is not able to say with certainty that a treatment caused an angler to purchase a fishing license. It is impossible to know what the purchase rate for any of the participants would be if they would have not participated. Without further research, we are not able to determine the angler s motivation when purchasing a fishing license; it is possible that a treatment caused them to become interested in fishing. However, it is also the case that an angler could have made the decision to purchase a license before they decided to enter a treatment, but the dates of actual purchase were not aligned with the decision to make that purchase. While the results presented here are limited to suggesting that entering a treatment and purchasing a fishing license are merely correlated, the resulting analysis of purchasing history will be able to describe differences in fishing license purchasing patterns both before and after treatments. Furthermore, differences in demographics will also account for differences in license purchasing history, as some groups are inherently more likely to purchase a license than others. Methodology Page 6

Results Results for each treatment are presented in the sub-sections below, with table numbers corresponding to the treatment. All treatments took place over multiple years; as such, all demographic data represent the individual at their time of treatment. Similarly, R3 variables are calculated based on their license buying history in the years preceding the angler s treatment date. 1 Treatment 1: Contests Contest entry forms (paper forms) were completed by individuals at shows and events where PFBC had a booth. The recruitment of potential anglers was the primary objective of this passive treatment, seeking to obtain email addresses of those who had not previously purchased a fishing license from PFBC. These contests provided fishing or boating prizes as an incentive to provide PFBC with email addresses and other information. The treatment dates used in this analysis represent when the individuals were entered into PFBC s system, not the date when the entries were collected. Five contests are represented here: Boat package Tracker Boat (2016); Fall into Fishing (2016); PSU Winter Sports (2016); Gone Fishing, PA! (2017); and Blue-White Game (2017). Data Overview The following data were collected from individuals participating in these contests: last name, first name, date of birth, and address, including street address, city, state, and ZIP code. Data fields included in the contest registrations follow: last name; first name; date of birth; and home address, including street address, city, state, and ZIP code. Table 1.0: Data Completeness Overview (Contests) Status Count Percent Complete 10,076 95.9% Missing DOB 0 0.0% Missing Address 430 4.1% Missing DOB and Address 0 0.0% Total Observations 10,506 100.0% With 10,506 contest entries made by 10,065 individuals, many of the participants in is treatment entered the contests only once. About one-third (3,596 of 10,506) of the individuals in this treatment 1 For example, if an angler entered a treatment in year Y, and had no history of purchases ranging back to year Y-5, then they would be considered recruited. In this case, Y is a variable describing the year that an individual entered the treatment and can be different for all individuals. This is different from traditional R3 definitions which use calendar or license years to define an individual s status. Results: Contests Page 7

were able to be matched to a record in PALS. This is likely a reflection of the treatment s goal of reaching potential anglers who had not previously purchased a fishing license from PFBC. Table 1.1: PALS Record Identification (Contests) Status Count Observations 10,506 Unique participants 10,065 Matched to PALS 3,596 Demographics The majority of those in the treatment group (60.4%) were between the ages of 35 and 64 years old when entering the treatment. More than one-quarter (27.9%) of the treatment group is younger than 35 while just 11.7% are 65 or older. Those in this treatment group are more likely to fall into the 35 to 64 year age group than the general PALS population (60.4% vs 51.6%). Because contest registration data included date of birth, the data presented below accurately represents the entire treatment, with the exception of 87 individuals who choose not to include this information. Table 1.2: Age Category (Contests) Age Group Treatment PALS (2017) Count Percent Count Percent 15 or younger 2 0.0% 30,790 3.1% 16 to 24 years 1,092 10.9% 136,938 13.9% 25 to 34 years 1,697 17.0% 158,819 16.1% 35 to 44 years 1,610 16.1% 152,113 15.4% 45 to 54 years 2,206 22.1% 164,101 16.7% 55 to 64 years 2,219 22.2% 153,289 15.6% 65 to 74 years 1,039 10.4% 156,500 15.9% 75 to 84 years 108 1.1% 30,823 3.1% 85 years and older 5 0.1% 2,125 0.2% Total 9,978 100% 985,498 100% Because registration data for this treatment did not include a field for gender, the table below represents the gender of those individuals whose records were identified in the PALS data set. Accordingly, Table 1.3 below should be understood as the gender makeup of the sample of treatment individuals who are also license purchasers. Table 1.3: Gender (Contests) Gender Treatment PALS- 2017 Count Percent Count Percent Male 2,591 72.1% 792,895 81.3% Female 1,003 27.9% 191,910 18.7% Total 3,594 100% 984,805 100% Results: Contests Page 8

License Status at Time of Treatment The following table indicates the license purchasing history of treatment group members at the time of the treatment. Many individuals within this treatment actively held a valid license prior to the treatment date. One-fourth of the individuals within this treatment had purchased a license from PFBC in one of the years preceding the treatment date, ranging back to 2008. However, about two-thirds of all individuals within the treatment frame were not found in the PALS data set. There were 979 (9.7%) anglers who purchased a PA fishing license prior to entering the treatment within the same license year that the treatment occurred. The portion of these individuals who were recruited or reactivated are represented below, in the "Prior to treatment..." line within table 1.6. The remainder of individuals who held a valid fishing license at the time of treatment were retained anglers. At the time of treatment, there were 9,086 individuals who were not identified as having purchased a PA fishing license for the treatment year. Some of these individuals (2,501) had purchased a license at some point since 2008 and are categorized as prior customers. Others (6,585) had no known history of purchases in the PALS system dating back through 2008 and are categorized as potential customers. Table 1.4: License Status at Time of Treatment (Contests) Held Valid PA Fishing License? Count Percent Yes 979 9.7% No 9,086 90.3% - Prior Customers 2,501 24.8% - Potential Customers 6,585 65.4% Total 10,065 100% Recruitment, Reactivation, and Retention The following table focuses on only those that purchased a fishing license in the year of their treatment, or up to 18 months after the individual s treatment date. The table indicates the status of each angler s PA fishing license in the previous year compared to the treatment year. Most anglers (84.8%) were retained, having purchased a PA fishing license in the year prior to treatment. Of those licensed anglers who did not hold a license in the year prior to treatment, a slightly greater proportion of anglers had purchased a license in the past five years, designating them as reactivated (8.0%), while only (7.2%) were recruited, having not purchased a license within the five previous years. In all, 2,535 (25.2%) of the 10,065 treated individuals purchased a license in their treatment year, or up to 18 months after their treatment date. Of the 9,086 participants who had either lapsed or had never purchased a PA fishing license before, 451 (5.0%) were converted to active customers following their treatment. Results: Contests Page 9

Table 1.5: R3 (Contests) R3 Status Treatment PALS (2017) Count Percent Count Percent Retained 2,084 84.8% 667,297 67.7% Reactivated 225 8.0% 124,667 12.7% - S.T. Lapse 168 6.2% 95,113 9.7% - L.T. Lapse 57 1.8% 29,554 3.0% Recruited 226 7.2% 193,534 19.6% Total 2,535 100% 985,498 100% Of those anglers designated as reactivated or recruited, many purchased their PA fishing license prior to entering the treatment within the same license year (37.7%). This suggests that the treatment appealed to many anglers who had recently purchased a license, but also motivated other anglers to purchase a license (62.3%). Table 1.6: Purchase Timing of Reactivated & Recruited Customers (Contests) Reactivated Anglers Recruited Anglers Count Percent Count Percent Prior to treatment, within the same license year 84 37.3% 86 38.1% After the treatment date, within 1 month of the 21 9.3% 20 8.8% treatment date 1 month after the treatment date, to less than 6 106 47.1% 98 43.4% months following the treatment date 6 months after the treatment date, to less than 12 14 6.2% 22 9.7% months following the treatment date 12 months after the treatment date, to 18 months 0* 0.0%* 0* 0.0%* following the treatment date Total 225 100% 226 100% Within the treatment sample frame, the individuals who entered contests had significantly less churn (13.2% - 21.2%) than the greater license buying population (37.2% - 40.2%). There is a general trend of less churn in the more recent years among those in the treatment group. This could signal increased avidity, and that more avid anglers are more likely to participate in contests like these. The rate of reactivated and recruited individuals did not outpace the rate of churn for the treatment year. Results: Contests Page 10

Table 1.7: Churn Rate (Contests) Year Treatment PALS Customers Churn Customers Churn 2008 1,331 n/a 747,280 n/a 2009 1,413 20.8% 793,663 38.9% 2010 1,695 17.1% 846,754 39.2% 2011 1,676 21.2% 827,616 40.2% 2012 1,822 17.9% 887,480 37.7% 2013 1,954 17.1% 912,157 39.1% 2014 2,120 16.2% 945,863 37.4% 2015 2,264 15.1% 990,620 37.2% 2016 2,425 13.2% 998,950 38.6% 2017 2,486 13.2% 985,498 39.1% The following table indicates that 44.2% of the 678 anglers who entered the treatment in 2016 had purchased a fishing license from PFBC in 2011, and that 64.6% purchased a license in the year following the treatment, 2017. Several anglers appear to have entered contests in years that they did not hold valid fishing licenses; this is why the treatment year percent is less than 100%. This could simply be due to an inability to precisely match customers to their PALS records based on information like name, date of birth and address alone. Table 1.8: Five Year Purchasing Pattern (Contests) 2016 2017 Year Treatment PALS Treatment PALS Count Percent Count Percent Count Percent Count Percent 2011 300 44.2% 446,277 44.7% - - - - 2012 333 49.1% 490,616 49.1% 1489 51.0% 463,585 47.0% 2013 347 51.2% 530,584 53.1% 1607 55.1% 498,170 50.6% 2014 373 55.0% 583,946 58.5% 1748 59.9% 540,120 54.8% 2015 421 62.1% 663,476 66.4% 1844 63.2% 602,392 61.1% 2016 444 65.5% 998,950 100.0% 1981 67.9% 667,297 67.7% 2017 438 64.6% 667,297 66.8% 2049 70.2% 985,498 100.0% Total 678 100.0% - - 2918 100% - - Results: Contests Page 11

Treatment 2: Angler Awards Anglers submitted information to PFBC about fish they had caught, and those anglers whose fish weighed more than a given qualifying weight received an angler award patch and certificate in return. The main objectives of this passive treatment were to retain anglers by providing an incentive to fish, and to recruit new anglers by providing acknowledgement of their first catch. 2 Data Overview The following data were collected from individuals at the time of entry: last name, first name, age at the time of their catch, and address, including street address, city, state, and ZIP code. Date of birth was not collected for this treatment. Table 2.0: Data Completeness Overview (Angler Awards) Status Count Percent Complete 0 0% Missing DOB 5,020 100% Missing Address 0 0% Missing DOB and Address 0 0% Total Unique Observations 5,020 100% About one-third of the observations were individuals who had previously participated in this treatment. Because this treatment recorded an angler s age instead of date of birth, the process by which treatment individuals were matched to PALS records was negatively affected. The variables used to match participants with PALS records included name (first and last) and address, which resulted in relatively fewer PALS matches than other treatments. The PALS records were identified for more than half of the unique participants (1,802 of 3,375). Table 2.1: PALS Record Identification (Angler Awards) Status Count Observations 5,020 Unique participants 3,375 Matched to PALS 1,802 Demographics Table 2.2 below presents age data derived from the variable age at the time of catch that participants noted in their registration. Almost all (98.5%) of those in the treatment had a valid age at the time of catch observation, while about 1.5% were missing this variable. The age distribution of this treatment closely follows the distribution found in the 2017 PALS population, although there were fewer anglers older than 65 in this treatment (11.1%) than in PALS (19.2%). 2 Because this treatment included the incentivization of an angler s first catch, this treatment drew participation from many anglers ages 15 and younger. However, because PFBC does not require anglers ages 15 and younger to purchase a fishing license, the results presented here only consider anglers ages 16 and older. Results: Angler Awards Page 12

Table 2.2: Age Category (Angler Awards) Age Group Treatment PALS- 2017 Count Percent Count Percent 15 or younger 0 0% 30,790 3.1% 16 to 24 years 425 12.8% 136,938 13.9% 25 to 34 years 586 17.6% 158,819 16.1% 35 to 44 years 614 18.5% 152,113 15.4% 45 to 54 years 708 21.3% 164,101 16.7% 55 to 64 years 623 18.7% 153,289 15.6% 65 to 74 years 315 9.5% 156,500 15.9% 75 to 84 years 47 1.4% 30,823 3.1% 85 years and older 8 0.2% 2,125 0.2% Total 3,326 100% 985,498 100% Because gender was not recorded with treatment registration, the table below presents gender data only for those participants whose PALS records were identified. A smaller proportion of women were represented within this treatment compared to general license sales. Table 2.3: Gender (Angler Awards) Gender Treatment PALS (2017) Count Percent Count Percent Male 1,595 88.5% 792,895 81.3% Female 207 11.5% 191,910 18.7% Total 1,802 100% 984,805 100% License Status at Time of Treatment Many individuals within this treatment had not purchased a fishing license from PFBC within the treatment year prior to entering the treatment. Only 10.6% of the unlicensed individuals within this treatment had purchased a license from PFBC in one of the years preceding the treatment date, ranging back to 2008. About two-thirds (47.8%) of all individuals within the treatment group were not found in the PALS data set. This could be due to the difficulty of matching customers to their PALS records based on name and address alone, as date of birth was not collected for this treatment. There were 1,403 anglers who purchased a PA fishing license prior to entering the treatment within the same license year that the treatment occurred. The portion of these individuals who were recruited or reactivated are indicated below in the "Prior to treatment..." line within Table 2.6. The remainder of individuals who held a valid fishing license at the time of treatment were retained anglers. At the time of treatment, there were 1,613 individuals who had not purchased a PA fishing license for the year in which the treatment occurred. Some of these individuals (359) had purchased a license at some point in the past, categorized as prior customers, while others (1,613) had no known history of purchases in the PALS system, categorized as potential customers. Results: Angler Awards Page 13

Table 2.4: License Status at Time of Treatment (Angler Awards) Held Valid PA Fishing License? Count Percent Yes 1,403 41.6% No 1,972 58.4% - Prior Customers 359 10.6% - Potential Customers 1,613 47.8% Total 3,375 100% Recruitment, Reactivation, and Retention The following table focuses on only those that purchased a fishing license in the year of their treatment and indicates the status of their PA fishing license in the previous year. Most of these anglers (81.4%) were retained, having purchased a PA fishing license in the year prior to treatment. A small portion of anglers had purchased a license in the past five years, designating them as reactivated (2.7%), while 4.9% were recruited, having not purchased a license within the five previous years. In all, 1,821 (54.0%) of the 3,375 treated individuals purchased a license in the treatment year, or up to 18 months following their treatment. Of the 3,403 anglers who had lapsed or had not previously purchased a PA fishing license at all, 140 (4.1%) were converted to active customers following the treatment. Table 2.5: R3 (Angler Awards) R3 Status Treatment PALS (2017) Count Percent Count Percent Retained 1,483 81.4% 667,297 67.7% Reactivated 50 2.7% 124,667 12.7% - S.T. Lapse 38 2.1% 95,113 9.7% - L.T. Lapse 12 0.7% 29,554 3.0% Recruited 90 4.9% 193,534 19.6% Total 1,821 100% 985,498 100% Of those anglers designated as reactivated or recruited, more than half (82.1%) purchased their PA fishing license prior to entering the treatment, within the same license year. Table 2.6: Purchase Timing of Reactivated & Recruited Customers (Angler Awards) Reactivated and Purchase Timing of Reactivated & Recruited Recruited Anglers Customers Count Percent Prior to treatment, within the same license year 115 82.1% After the treatment date, within 1 month of the treatment date 14 10.0% 1 month after the treatment date, to less than 6 months following the treatment date 1 0.7% 6 months after the treatment date, to less than 12 months following the treatment date 8 5.7% 12 months after the treatment date, to 18 months following the treatment date 2 1.4% Total 140 100% Results: Angler Awards Page 14

The rate of churn among these individuals is the lowest calculated for the six treatments analyzed, with the highest rate of churn still measuring below 10%. The low churn rate corresponds with the high retained rate in table 2.5. The rate of reactivated and recruited individuals kept pace with the churn rate for the treatment periods. Table 2.7: Churn Rate (Angler Awards) Year Treatment PALS Customers Churn Customers Churn 2008 1144 NA 747,280 NA 2009 1207 8.4% 793,663 38.9% 2010 1400 5.2% 846,754 39.2% 2011 1475 4.7% 827,616 40.2% 2012 1480 6.6% 887,480 37.7% 2013 1524 5.4% 912,157 39.1% 2014 1517 6.4% 945,863 37.4% 2015 1518 6.7% 990,620 37.2% 2016 1494 6.4% 998,950 38.6% 2017 1422 8.4% 985,498 39.1% Here, the data show a purchasing pattern for every treatment year. In the years preceding the treatment, the purchase rates for treated individuals are lower than the rates during the treatment year or in those years after. After participating in the Angler Awards treatment, many individuals experienced an increased rate of retention, renewing licenses more often than prior to the treatment. Table 2.8: Five Year Purchasing Pattern (Angler Awards) 2010 2011 Year Treatment PALS (2010) Treatment PALS (2011) Count Percent Count Percent Count Percent Count Percent 2008 205 73.0% 440,358 52.0% 166 65.9% 400,526 48.4% 2009 214 76.2% 506,006 59.8% 181 71.8% 449,523 54.3% 2010 274 97.5% 846,754 100.0% 216 85.7% 537,849 65.0% 2011 256 91.1% 537,849 63.5% 246 97.6% 827,616 100.0% 2012 238 84.7% 506,135 59.8% 220 87.3% 551,071 66.6% 2013 227 80.8% 476,259 56.2% 217 86.1% 507,741 61.3% 2014 227 80.8% 457,349 54.0% 198 78.6% 481,555 58.2% 2015 216 76.9% 443,359 52.4% 196 77.8% 462,766 55.9% 2016 - - - - 191 75.8% 446,277 53.9% Total 281 100% - - 252 100.0% - - Results: Angler Awards Page 15

Table 2.8.1: Five Year Purchasing Pattern Continued (Angler Awards) 2012 2013 Year Treatment PALS (2012) Treatment PALS (2013) Count Percent Count Percent Count Percent Count Percent 2008 178 65.7% 392,445 44.2% 169 66.3% 376,496 41.3% 2009 190 70.1% 434,729 49.0% 176 69.0% 414,080 45.4% 2010 208 76.8% 506,135 57.0% 199 78.0% 476,259 52.2% 2011 237 87.5% 550,463 62.0% 210 82.4% 507,741 55.7% 2012 260 95.9% 887,480 100.0% 215 84.3% 583,377 64.0% 2013 240 88.6% 583,377 65.7% 242 94.9% 912,157 100.0% 2014 225 83.0% 541,183 61.0% 220 86.3% 608,984 66.8% 2015 220 81.2% 512,494 57.7% 218 85.5% 564,432 61.9% 2016 209 77.1% 490,616 55.3% 216 84.7% 530,584 58.2% 2017 202 74.5% 463,585 52.2% 201 78.8% 498,170 54.6% Total 271 100% - - 255 100% - - Table 2.8.2: Five Year Purchasing Pattern Continued (Angler Awards) 2014 2015 Year Treatment PALS (2014) Treatment PALS (2015) Count Percent Count Percent Count Percent Count Percent 2009 167 57.6% 400,997 42.4% - - - - 2010 197 67.9% 457,349 48.4% 153 66.5% 443,359 44.8% 2011 206 71.0% 481,555 50.9% 160 69.6% 462,766 46.7% 2012 218 75.2% 541,183 57.2% 164 71.3% 512,494 51.7% 2013 238 82.1% 608,984 64.4% 185 80.4% 564,432 57.0% 2014 278 95.9% 945,863 100.0% 191 83.0% 634,743 64.1% 2015 255 87.9% 634,743 67.1% 220 95.7% 990,620 100.0% 2016 245 84.5% 583,946 61.7% 205 89.1% 663,476 67.0% 2017 234 80.7% 540,120 57.1% 191 83.0% 602,392 60.8% Total 290 100% - - 230 100% - - Table 2.8.3: Five Year Purchasing Pattern Continued (Angler Awards) 2016 2017 Year Treatment PALS (2016) Treatment PALS (2017) Count Percent Count Percent Count Percent Count Percent 2011 130 71.8% 446,277 44.7% - - - - 2012 135 74.6% 490,616 49.1% 30 71.4% 463,585 47.0% 2013 145 80.1% 530,584 53.1% 30 71.4% 498,170 50.6% 2014 148 81.8% 583,946 58.5% 30 71.4% 540,120 54.8% 2015 158 87.3% 663,476 66.4% 35 83.3% 602,392 61.1% 2016 175 96.7% 998,950 100.0% 36 85.7% 667,297 67.7% 2017 161 89.0% 667,297 66.8% 40 95.2% 985,498 100.0% Total 181 100% - - 42 100% - - Results: Angler Awards Page 16

Treatment 3: Family Fishing Program (All Types) This treatment was a hands-on, active education program targeting families and providing basic fishing skills as well as the opportunity to fish. This treatment emphasized recruiting and reactivating anglers by increasing the skills, awareness of fishing opportunities, and confidence of participating families. Fishing license requirements were waived for participants ages 16 and older. Data Overview The following data were collected from individuals at the time of entry: last name, first name, gender, date of birth, and address, including street address, city, state, and ZIP code. There were a total of 2,751 observations within this treatment. Table 3.0: Data Completeness Overview (Family Fishing) Status Count Percent Complete 2,383 86.6% Missing DOB 122 4.4% Missing Address 267 9.7% Missing DOB and Address 21 0.8% Total Observations 2,751 100.0% Many individuals who attended the Family Fishing treatments were not able to be matched to PALS records, but this is likely due to license requirements having been waived for participants age 16 and older, while anglers younger than 16 are not required to have a PA fishing license in order to fish. Table 3.1: PALS Record Identification (Family Fishing) Status Count Observations 2,751 Unique participants 2,671 Matched to PALS 830 Demographics Age and gender were both recorded in the registration data set for this treatment. There were relatively few observations missing date of birth records (Table 3.0), but only 803 of 2,751 treatment individuals had recorded their gender information. Over two-thirds of those in the treatment group (68.3%) were between the ages of 35 and 64 years old, while anglers younger than 35 years old accounted for 23.6%. Results: Family Fishing Page 17

Table 3.2: Age Category (Family Fishing) Age Group Treatment PALS (2017) Count Percent Count Percent 15 or younger 81 3.3% 30,790 3.1% 16 to 24 years 125 5.0% 136,938 13.9% 25 to 34 years 382 15.3% 158,819 16.1% 35 to 44 years 944 37.9% 152,113 15.4% 45 to 54 years 501 20.1% 164,101 16.7% 55 to 64 years 256 10.3% 153,289 15.6% 65 to 74 years 158 6.3% 156,500 15.9% 75 to 84 years 30 1.2% 30,823 3.1% 85 years and older 12 0.5% 2,125 0.2% Total 2,489 100% 985,498 100% Additional PALS data was not appended to the treatment data for this table so that these results more accurately represent the gender composition of the treatment group. A greater proportion of women were represented within this treatment (54.2%) compared to general license sales (18.7%). Table 3.3: Gender (Family Fishing) Gender Treatment PALS- 2017 Count Percent Count Percent Male 368 45.8% 792,895 81.3% Female 435 54.2% 191,910 18.7% Total 803 100% 984,805 100% License Status at Time of Treatment At the time of treatment, there were 2,369 individuals who had not purchased a PA fishing license for the year in which the treatment occurred. Some of these individuals (323) had purchased a license at some point in the past and are categorized as prior customers. Others (2,046) had no known history of purchases in the PALS system and are categorized as potential customers. Table 3.4: License Status at Time of Treatment (Family Fishing) Held Valid PA Fishing License? Count Percent Yes 302 11.3% No 2,369 88.7% - Prior Customers 323 12.1% - Potential Customers 2,046 76.6% Total 2,671 100% Recruitment, Reactivation, and Retention The following table focuses only on those who held a valid fishing license in the year of the treatment and indicates the status of each angler s PA fishing license in the year before the treatment year. Most Results: Family Fishing Page 18